The non-stationary interference with the analysis parameters of short-term heart rate variability (HRV) has been studied. The results show that although all considered parameters fluctuate with the lapse of time, only HFnorm and sample entropy (SE) correlate significantly with non-stationary interference. Therefore, the low frequency fluctuations of HRV contain not only the non-stationary interference, but also the intrinsic cardiac dynamic characteristics. Specially, the results of the base-scale entropy (BE) analysis of three groups of people, and of three kinds of noises are contrasted. It is shown that, at the expense of outstanding anti-nonstationarity, BE fails to depict the long range correlation of HRV.